Proc. Natl. Acad. Sci. USA Vol. 76, No. 7, pp. 3083-3086, July 1979 Biochemistry

L-Glucosylceramide: Synthesis, properties, and resistance to catabolism by glucocerebrosidase in vitro (glucocerebroside/Gaucher disease/animal model/chemical sphingolipidosis) ANDREW E. GAL, PETER G. PENTCHEV, JANICE M. MASSEY, AND ROSCOE 0. BRADY Developmental and Metabolic Neurology Branch, National Institutes of Neurological and Communicative Disorders and Stroke, National Institutes of Health, Bethesda, Maryland 20205 Contributed by Roscoe 0. Brady, March 16, 1979

ABSTRACT Procedures for the synthesis and radioactive activity by the administration of conduritol-3-epoxide to induce labeling of L-glucosylceramide are described. This compound a syndrome resembling Gaucher disease in mice (6). However, is a stereoisomeric analogue of D-glucosylceramide which oc- the quantity of glucosylceramide that accumulated in liver and curs in nature and accumulates in pathological quantity in the 2- to 3-fold than that in normal mice organs and tissues of patients with Gaucher disease. The prop- spleen was only greater erties of L-glucosylceramide that have been examined so far and is therefore far below that found in patients with Gaucher have been found to be indistinguishable from those of the nat- disease (4). In addition, the typical Gaucher bodies or tubular urally occurring glycolipid. However, L-glucosylceramide is structures characteristic of the disorder were not observed in completely refractory to enzymatic hydrolysis by purified pla- spleen, liver, or bone marrow of mice treated with this reagent cental glucocerebrosidase and enzyme(s) present in whole tissue (7). extracts. It is anticipated that L-glucosylceramide will be a disease model at uniquely useful substance for exploring pathogenetic processes Because of the lack of a suitable Gaucher in animal analogues of Gaucher disease. the enzyme level, investigators have attempted with little success, to produce Gaucher disease-like lesions by the ad- The pathogenesis of most lysosomal storage disorders is poorly ministration (intraperitoneally or orally) of high levels of understood at this time although clear definitions of the primary galactosylceramide to rats and rabbits (8). When radioactive enzymatic lesions often exist (1). Because of this situation, in- glucosylceramide was administered intravenously to rats, it was formation relating to the pathologic manifestations of these rapidly cleared from the plasma and taken up by the reticulo- diseases has depended primarily on descriptions of clinical endothelial system where it was quickly catabolized, and no courses in patients afflicted with these disorders and on the accumulation could be demonstrated (unpublished data). The findings at postmortem examinations. The pathogenetic basis pronounced susceptibility of glucosylceramide to normal cat- of the anatomical and physiological alterations of these disorders abolic processes pointed to the desirability of attempting to has been largely confined to speculation. Gaucher disease is a prepare a substrate analogue with the chemical and physical classic example of this dilemma. This disorder is caused by a properties of the natural glycolipid but with the added feature deficiency of glucocerebroside-3-glucosidase activity in the of its being resistant to enzymatic hydrolysis. It was considered tissues of affected individuals (2). It has recently been shown likely that glucosylceramide containing L-glucose instead of in a patient with non-neuropathic form of the disorder (type the naturally occurring D-glucose might be useful in this re- I) that this deficiency arises from a structural mutation in the gard. enzyme that results in markedly decreased hydrolysis of glu- We describe here the syntheses of 1-O-3-D-glucosyl-N- cosylceramide without apparent effect on the affinity of the palmitoyl-DL-sphingosine and 1-O-3-L-glucosyl-N-palmi- enzyme for the substrate (3). The molecular bases that distin- toyl-DL-sphingosine and compare the chemical and physical guish the two neuropathic forms of Gaucher diseases (types II properties of the two stereoisomers. In addition, the catabolic and III) from type I Gaucher disease are not known. In addition, inertness of L-glucosylceramide in vitro is demonstrated. the pathophysiologic consequences of the accumulation of glucosylceramide are poorly understood. For example, it has MATERIALS AND METHODS been observed that there is a wide variation in the quantity of Materials glucosylceramide in the liver of patients with Gaucher disease Silica gel 60 (E. Merck) plates were used for thin-layer chro- and that there often is little or no correlation of this accumu- matography. The migration of glycolipids was visualized by lation with (i) the level of residual glucocerebrosidase activity charring with ammonium bisulfate (9). The silica gel used for or (ii) the severity of the clinical manifestations (4). column chromatography was Bio-Sil HA (minus 35 mesh; The availability of an experimental model of Gaucher disease Bio-Rad) that was activated prior to use by heating at 1100C seems to be a paramount requirement for answering many of for 12 hr. Hydrogen bromide in acetic acid was obtained from these puzzling questions. Although a strain of mutant BALB/c Koch-Light Labs (Colnbrook, England). Radioactive mea- mice with hepatic and splenic accumulation of glucocerebroside surements were carried out in a Searle mark III liquid scintil- has recently been reported, the clinical manifestations of the lation system using Aquasol (New England Nuclear). 3-0- disease in these mice are not typical of Gaucher disease (5). Benzoyl-N-palmitoyl-DL-sphingosine, mp 77-78, and 1-0- Investigators have attempted to inhibit glucocerebrosidase 3-D-glucopyranosyl-N-[1-14Clstearoyl-DL-sphingosine (0.315 mCi/mmol; 1 Ci = 3.7 X 1010 becquerels) were kindly provided The publication costs of this article were defrayed in part by page charge payment. This article must therefore be hereby marked "ad- by D. Shapiro. L-[1-14C]Glucose was purchased from New vertisement" in accordance with 18 U. S. C. §1734 solely to indicate England Nuclear. Unlabeled L-glucose was obtained from this fact. Sigma. Radioscans were made with a series 6000 Varian 3083 Downloaded by guest on September 24, 2021 3084 Biochemistry: Gal et al. Proc. Natl. Acad. Sci. USA 76 (1979) (Berthold) radioscanner. Melting points were taken on a Thomas-Hoover melting point apparatus and are reported corrected. Synthesis of glucocerebrosides 1,2,3,4,6-Penta-O-acetyl-,-L-glucopyranoside. A mixture of 5.4 g (30 mmol) of L-glucose, 30 ml of acetic anhydride, and 2.5 g of previously melted sodium acetate was heated at 950C for 2 hr. The mixture was decomposed with 12.0 g of ice water, stirred for 3 hr at 00C, and kept for 1 day at 250C. The pre- cipitate was filtered and washed with 600 ml of water, dried on the filter, and washed with 60 ml of benzene/hexane, 1:1 (vol/vol). The product was recrystallized from ethanol (600 ml). The yield was 5.8 g (50%) having a melting point of 130-1310C. On thin-layer chromatography in cyclohexane/benzene/iso- propylether/pyridine, 115:20:20:25 (vol/vol), the pentaacetate had RF 0.33. Pentaacetyl-D-glucopyranoside had the same RF value in this system: 2,3,4,6-Tetra-O-acetyl-a-L-glucopyranosyl Bromide. A A B c solution of 3.9 g (10 mmol) of L-pentaacetylglucose in 20 ml of 45% (wt/vol) hydrogen bromide in acetic acid was kept at 4VC FIG. 1. Radioscan of a thin-layer chromatogram of 1-0; for 18 hr and subsequently at 250C for 4 hr. The mixture was f3-L-glucopyranosyl N-[1-14C]palmitoyl-DL-sphingosine. Lanes: A, then decomposed by the addition of 16 ml of ice water. The purified ceramide standard; B, L-[1-'4CJglucosyl ceramide; C, un- product was extracted with 300 ml of ether, and this solution reacted ceramide from the reaction mixture. The solvent system was was washed at 00C with three consecutive 400-ml portions of chloroform/methanol/water, 40:10:1 (vol/vol). a 0.5 M potassium bicarbonate. The product was dried over anhydrous calcium chloride, the solvent was evaporated, and were identical to those of the L isomer. From this reaction 300 the residue was recrystallized from 20 ml of isopropyl ether. mg (56%) of N-palmitoylsphingosine was recovered. The yield was 1.95 g (47%) with a melting point of 88-890C. 1-O-3-L-Glucopyranosyl-N{1-14C palmitoyl-DL-sphingo- Thin-layer chromatography in the above described system gave sine. This was prepared according to the method for the cor- an RF 0.47 for the bromide. The melting point and RF for the D-isomer were the same. 1-O-,-.L-GIucopyranosyI-.N-palmitoyl-DL-sphingosine. 3-O-Benzoyl-N-palmitoylsphingosine (642 mg; 1 mmol) was heated in 30 ml of nitromethane and 30 ml of benzene until 15 ml of the benzene had been distilled off. The mixture was cooled and 505 mg (2 mmol) of mercuric cyanide and 411 mg (1 mmol) of 2,3,4,6-tetra-0-acetyl-a-L-glucopyranosyl bromide was added. The mixture was heated with stirring at 720C for 2 days. After cooling, 200 ml of ether was added and the solu- tion was repeatedly washed with 0.5 M sodium bicarbonate until no mercuric ions were detected with a sodium sulfhydrate solution. The solvents were removed under reduced pressure and the residue was dissolved in 50 ml of methanol. Eight milliliters of 0.5 M sodium methylate in methanol was added; after 24 hr at room temperature, the mixture was neutralized with 0.23 ml of glacial acetic acid. To this solution were added 120 ml of chloroform and 36 ml of water. The suspension was shaken and the lower phase was washed once with the theo- retical upper phase (10) and evaporated; the residue was dis- solved in 400 ml of chloroform/methanol, 50:1 (vol/vol). The solution was chromatographed on a 50-g column of silic acid that was eluted with three chloroform/methanol mixtures: 1 liter of 50:1 (vol/vol), 600 ml of 30:1, and 300 ml of 9:1. The elution of glucocerebroside was monitored by thin-layer chromatography. The yield of L-glucosylceramide was 270 mg (39%). The product was crystallized from methanol (70 ml) and had a melting point of 182-1830C. The identity of the product was confirmed by elemental analysis. Calculated for C4oH-nNO8 (700.07): C, 68.63; H, 11.09; N, 2.00. Found: C, E 69.28; H, 11.24; N, 2.04. Thin-layer chromatography in chlo- A B C D an value roform/methanol/water, 40:10:1 (vol/vol), gave RF FIG. 2. Thin-layer chromatogram of synthetic D- and L-gluco- for L-gluco-cerebroside of 0.55 (Fig. 1). In a solvent system of cerebroside. Lanes: A, galactosylglucosylceramide; B, D-galactosyl- chloroform/methanol/water, 65:25:4, the RF value was 0.8 ceramide; C, L-glucosylceramide; D, D-glucosylceramide; E, L-Il- (Fig. 2). The melting point and RF values for D-glucosyl-N- 14Cjglucosylceramide. The solvent system was chloroform/metha- Palinitoylsphingosine prepared by the above described method nol/water, 65:25:4 (vol/vol). Downloaded by guest on September 24, 2021 Biochemistry: Gal et al. Proc. Natl. Acad. Sci. USA 76 (1979) 3085

wp .,II$ responding nonradioactive cerebroside described abov. The pI- .I.. first step of the synthesis used 540.5 mg (3 mmol) of L-[1- 14C]glucose (1.17 mCi/mmol). The yield of tetraacetyl-L-[1- 14C]glucopyranosyl bromide was 222 mg (0.55 mmol). The compound appeared to be homogeneous by thin-layer chro- matography, with a single corresponding peak of radioactivity (Fig. 1). 1-O-f-D-Glucopyranosyl-N-palmitoyl-DL-sphingosine. The starting material was purified tetraacetyl-D-glucopyranosyl bromide (11), 62 mg (0.15 mmol); all other reagents used were 1/10th the quantities used for the synthesis of the L isomer. The yield was 18 mg; the product melted at 182-183°C. Assay of glucocerebrosidase activity The enzyme source consisted of either highly purified human placental glucocerebrosidase (12) or whole mouse tissue extracts prepared as 20% (wt/vol) homogenates in water. Enzymatic activity was measured in 50 mM potassium phosphate, pH 5.9 in 0.12% Cutscum/0.05% crude sodium taurocholate (Difco) A B C D a final volume of 0.2 ml. The reaction was initiated by the ad- dition of 54 nmol of D-[1-'4C]glucosylceramide as a 5-,ul sample in a solution of sodium taurocholate. After incubation for 1 hr FIG. 3. Incubation (24 hr) of D- and L-glucocerebroside with at 370C, the reaction was stopped by the addition of 1.0 ml of 10,000 units of purified human placental glucocerebrosidase. The cooled bovine serum albumin (10 mg/ml) followed by 0.1 ml reaction was stopped by the addition of chloroform/methanol, 2:1 trichloroacetic acid (1 g/ml, aqueous solution). The mixture (vol/vol); after partitioning, the contents of the lower phase were D- was swirled and then centrifuged at 3000 X g for 10 min. The analyzed by thin-layer chromatography. Lanes A and B represent and L-glucosylceramide, respectively, in the absence of glucocere- enzymatically liberated [1-14C]glucose in the acid-soluble su- brosidase. Lanes C and D represent D- and L-glucosylceramide in the pernatant was measured by liquid scintillation spectroscopy. presence of glucocerebrosidase. The dark spot in the middle of the plate corresponds to the RF value for glucosylceramide standard. The RESULTS dark spot at the top of lane C corresponds to ceramide. The other Synthesis of Glucocerebroside Analogues. 1,2,3,4,6- lipids seen on the plate originate from the partially purified sodium Penta-O-acetyl-fl-L-glucopyranose and 2,3,4,6-tetra-0-ace- taurocholate used in the assay. tyl-a-L-glucopyranosyl bromide were prepared in good yields by a modification of the method of Potter et al. (13). The cosylceramide standard. There was no indication of the for- melting points of these compounds were the same as those of mation of a labeled reaction product in the region of psychosine the D-isomers. The syntheses of the D and L glucosylceramides (Fig. 4). were based on the method of Shapiro (14). The melting points and RF values for the stereoisomeric cerebrosides were iden- DISCUSSION tical. L-[1-'4C]Glucosylceramide was prepared on a smaller The stereochemistry of the D- and L-glucosylceramides is scale in good yield. somewhat complex. Due to the additional asymmetric nature Effect of Purified Glucocerebrosidase on the D- and L- of sphingosine, the two cerebrosides are not pure enantio- Glucosylceramides. No release of labeled hexose could be morphs. The synthetic sphingosine with which the respective detected when L-[1-'4C]glucosylceramide was incubated with cerebrosides were prepared was, itself, a racemic mixture of the enzyme under conditions routinely used for the hydrolysis D- and L-sphingosine. The conjugation to this racemic mixture of D-glucosylceramide. When the amount of enzyme and time of an additional optically active molecule (D- or L-glucose) of incubation were adjusted to achieve nearly complete hy- results in an equimolar mixture of diastereoisomers. Conse- drolysis of the D isomer, there was still no detectable hydrolysis quently, the stereochemical relationship between the synthetic of L-glucosylceramide (Fig. 3). Furthermore, the addition of the L isomer did not influence the hydrolysis of D-glucosyl- Table 1. Incubation of mouse tissue extracts with ceramide when the glycolipids were incubated together in D- or L-glucosylceramide equimolar concentration, indicating that the analogue con- Glucosylceramide hydrolyzed, taining L-glucose does not interact with the active site on the nmol/g (wet)/hr enzyme. Tissue D-Glucose isomer L-Glucose isomer Incubation of Tissue Extracts with D- and L-Glucosylcer- Liver 4800 0 amides. When the glycolipids were incubated with whole 5300 0 mouse tissue extracts under conditions that allow expression of Spleen Kidney 2200 0 most lysosomal hydrolases, there was substantial ,B-glucosidase Intestine 1500 0 activity observed with the D isomer but no detectable hydrolysis Thymus 4000 0 of the L isomer (Table 1). Because the normal mode of cleavage Brain 3000 0 of glucosidase did not appear to be functioning with L-glu- Lung 2300 0 cosylceramide, the possibility that the hydrolysis of the amide Plasma 5* 0 linkage resulted in the formation of glucosylsphingosine (psy- Frozen mouse tissues were thawed and homogenized by hand in 4 chosine) and free fatty acid was investigated. Examination of vol of distilled water with a tightly fitting all-glass homogenizer. A mixtures of tissue extracts incubated with L-[1-'4C]glucosyl- 25-pu aliquot of the whole homogenate was incubated with D- or L- ceramide by thin-layer chromatography revealed only a single [1-14C]glucosylceramide. radioactive area that corresponded with the migration of glu- * Shown as nmol/ml per hr. Downloaded by guest on September 24, 2021 3086 Biochemistry: Gal et al. Proc. Natl. Acad. Sci. USA 76 (1979) ..!in ... appearance of cellular responses such as lysosomal hypertrophy .. induced by the accumulation of L-glucosylceramide should be particularly instructive. The influence of the form in which the glycolipid is administered (free, associated with lipoproteins, .W. or incorporated into erythrocytes or leukocytes) may pro- iusp foundly affect the uptake of the glycolipid. Furthermore, it will be possible to carry out examinations of the potential intercel- lular exchange of L-glucosylceramide between cells of the re- ticuloendothelial system with parenchymal cells (e.g., Kupffer cells and hepatocytes) and the possibility of the excretion of this I lipid via the bile. It will also be of considerable interest to de- termine if L-glucosylceramide can enter anabolic pathways. The formation and disposition of such a "hybrid" synthetic natural product could provide considerable insight concerning the source, transport, and fate of complex sphingolipids. In sum, the numerous investigative possibilities concerning the metabolism of L-glucosylceramide might be thought of as A B representing a reversal of natural processes. Metabolic disorders occur because mutant enzymes are no longer capable of in- FIG. 4. Incubation of L-[1-C14Jglucosylceramide with a crude teracting properly with their normal substrates. One can now mouse liver homogenate. Lane A represents a mixture of the following attempt to alter physiological processes by presenting a "mu- standard glycolipids in order of diminishing RF values: glucosylcer- tant" substrate to normal enzymes. amide, galactosylceramide, galactosylglucosylceramide, galactosyl- galactosylglucosylceramide, and N-acetylgalactosaminylgalactosyl- 1. Brady, R. 0. (1976) Science 193,733-739. galactosylglucosylceramide (globoside). Lane B represents a chloro- 2. Brady, R. O., Kanfer, J. N. & Shapiro, D. (1965) Biochem. Bio- form/methanol extract of a 24-hr incubation of mouse whole liver phys. Res. Commun. 18,221-225. homogenate (20%, wt/vol) with L-[1-14C]glucosylceramide. The 3. Pentchev, P. G., Brady, R. O., Blair, H. E., Britton, D. E. & Sorrell, chromatogram was made on a silica plate in a solvent system of chloroform/methanol/water, 40:10:1 (vol/vol), scanned for radioac- S. H. (1978) Proc. Nati. Acad. Sci. USA 75,3970-3973. and subsequently visualized by charring. Psychosine(glucosyl) 4. Pentchev, P. G., Barranger, J. A., Gal, A. E., Furbish, F. S. & tivity, in and in Disease has been shown (15) to chromatograph with a RF value close to that Brady, R. 0. (1979) Glycoproteins Glycolipids of globoside. Processes, ACS Symposium Series, No. 80, ed. Walborg, E. F., Jr. (Am. Chem. Soc., Washington, DC), p. 156. 5. Pentchev, P. G., Boothe, A. D., Gal, A. E., Omodeo-Sale, F.& D- and L-glucosylceramide preparations consists of 50% en- Brady; R. 0. (1979) in Enzyme Therapy in Genetic Diseases, 2nd antiomorphism and 50% diastereoisomerism. Although a lack International Symposium, Hilton Head, South Carolina, March of sufficient material limited a rigorous analysis of the potential 3-7 (The National Foundation, New York), p. 106 (abstr.). chemical and physical differences due to diastereoisomerism, 6. Stephens, M. C., Bernatsky, A., Burachinsky, V., Legler, L. & the D- and L-glucosylceramide preparations appeared to be Kanfer, J. (1978) J. Neurochem. 30, 1023-1027. identical on the basis of their chemical properties described 7. Adachi, M. & Volk, B. W. (1977) Arch. Pathol. Lab. Med. 101, above. An additional example of their similarity was the iden- 255-259. tical migration of D- and L-glucosylceramides in thin-layer 8. Burton, R. M. & Sodd, M. A. (1969) Lipids 4,496-500. chromatographic systems (Fig. 2). The resolution obtained by 9. Gal, A. E. (1968) Anal. Biochem. 24,452-461. this system can be judged from the decidedly slower migration 10. Folch, J., Lees, M. & Stanley, G. H. S. (1957) J. Biol. Chem. 226, of galactosylceramide than its 4-epimeric analogue glucosyl- 497-509. 11. Gal, A. E., Pentchev, P. G. & Fash, F. J. (1976) Proc. Soc. Exp. ceramide. Biol. Med. 153,363-366. Because the refractoriness of L-glucosylceramide to enzy- 12. Furbish, F. S., Blair, H. E., Shiloach, J., Pentchev, P. G. & Brady, matic hydrolysis at both the glucosidic and amide linkages has R. 0. (1977) Proc. Nati. Acad. Sci. USA 74,3560-3563. been demonstrated in vitro, use of this glycolipid may be 13. Potter, A. L., Sowden, J. C., Hassid, W. Z. & Doudoroff, M. (1948) considered for a number of physiological and pathological in- J. Am. Chem. Soc. 70, 1751-1752. vestigations. The anticipated catabolic inertness of the L-hexosyl 14. Shapiro, D. (1969) in Chemistry of Sphingolipids (Hermann, glycolipid in vsvo may enable one to mimic the storage of D- Paris), p. 100. glucosylceramide that occurs in Gaucher disease with a high 15. Neskovic, N. M., Nussbaum, J. L. & Mandel, P. (1970) J. Chro- degree of fidelity. An investigation of the chronology of the matogr. 49,255-261. Downloaded by guest on September 24, 2021